Cache plugin registrant in test compiler (#185598)

Towards #69429 ("Make `flutter test` faster"). 

This change makes `TestCompiler` cache the plugin registrant instead
of regenerating it for each test file. Plugins are stable for the
lifetime of a single `flutter test`
run. However, the Dart version of test files might differ, so the cache
is keyed on the test file's
Dart version. 

**Benchmark**
Flutter foundation test suite (32 files, 220 tests), `flutter test -j
4`, 3 runs + 1 warmup:

|                                 | master | this PR    |
| ------------------------------- | ------ | ---------- |
| Avg compile per subsequent file | 68ms   | 55ms   |

That means the startup time for `flutter test` is reduced by ~13ms _per
test file_ on my machine. That quickly adds up and significantly
improves Flutter's TDD experience.

Results were generated with the following script:

<details><summary markdown="span">Benchmark</summary>

```bash
#!/usr/bin/env bash
# Reports per-run, for foundation suite, -j 4:
#   - wall      : total `flutter test` invocation time
#   - first     : compile time for the first test file (cache miss)
#   - subseq    : average compile time per subsequent file (cache hit / miss)
#   - total     : sum of all per-file compile times
#
# The "subseq" and "total" rows are the ones the PR moves; "wall" is dominated
# by test execution and is reported for completeness.

set -euo pipefail

FLUTTER_ROOT="${FLUTTER_ROOT:-$(pwd)}"
FLUTTER="$FLUTTER_ROOT/bin/flutter"
TARGET="$FLUTTER_ROOT/packages/flutter/test/foundation/"
RUNS="${RUNS:-3}"

cd "$FLUTTER_ROOT/packages/flutter"

bench_one() {
  local name="$1" log
  log=$(mktemp)
  local start end ms
  start=$(date +%s%N)
  "$FLUTTER" test -j 4 -v "$TARGET" >"$log" 2>&1
  end=$(date +%s%N)
  ms=$(( (end - start) / 1000000 ))

  local compiles n first sum rest_avg
  compiles=$(grep -oP 'Compiling .+ took \K\d+(?=ms)' "$log" || true)
  n=$(echo "$compiles" | grep -c . || true)
  first=$(echo "$compiles" | head -1)
  sum=$(echo "$compiles" | awk '{s+=$1} END{print s+0}')
  rest_avg=0
  if [[ "$n" -gt 1 ]]; then
    rest_avg=$(echo "$compiles" | tail -n +2 | awk '{s+=$1} END{print int(s/NR)}')
  fi
  printf "  %-8s wall=%5dms  first=%4dms  subseq=%3dms (n=%d)  total=%5dms\n" \
    "$name" "$ms" "$first" "$rest_avg" "$((n-1))" "$sum"
  rm -f "$log"
}

echo "Foundation suite (32 files, ~220 tests), -j 4, $RUNS runs + 1 warmup"
echo "Commit: $(cd "$FLUTTER_ROOT" && git rev-parse --short HEAD) ($(cd "$FLUTTER_ROOT" && git status --porcelain | wc -l) modified files)"
bench_one "warmup"
for i in $(seq 1 "$RUNS"); do
  bench_one "run_$i"
done
```

</details> 

## Pre-launch Checklist

- [x] I read the [Contributor Guide] and followed the process outlined
there for submitting PRs.
- [x] I read the [AI contribution guidelines] and understand my
responsibilities, or I am not using AI tools.
- [x] I read the [Tree Hygiene] wiki page, which explains my
responsibilities.
- [x] I read and followed the [Flutter Style Guide], including [Features
we expect every widget to implement].
- [x] I signed the [CLA].
- [x] I listed at least one issue that this PR fixes in the description
above.
- [x] I updated/added relevant documentation (doc comments with `///`).
- [x] I added new tests to check the change I am making, or this PR is
[test-exempt].
- [x] I followed the [breaking change policy] and added [Data Driven
Fixes] where supported.
- [ ] All existing and new tests are passing.